18 July 1964 Haemolytic Disease of Newborn-Walker et al. BRITISH 147

upon Tyne, for their co-operation; to Miss D. Purvis and Mrs. A.Hauxwell for the majority of the analyses; to Dr. T. R. C. Boydefor identifying the methaemalbumin ; and to Mrs. M. Malaghan forsecretarial assistance. Dr. S. M. Murray, of the Regional BloodTransfusion Centre, has taken an interest in the study throughoutand given much helpful advice, especially in the preparation of thisreport.

REFERENCES

Abbas, T. M., and Tovey, J. E. (1960). Brit. med. 7., 1, 476.Bangham, D. R., Hobbs, K. R., and Terry, R. J. (1958). Lancet, 2, 351.Beecham, C. T., Molthan, L., Boutwell, J., and Rohrbeck, C. W. (1962).

Amer. 7. Obstet. Gynec., 83, 1053.Bevis, D. C. A. (1950). Lancet, 2, 443.

(1952). Ibid., 1, 395.(1956). 7. Obstet. Gynaec. Brit. Emp., 63, 68.

Bourne, G. L. (1962). The Human Amnion and Chorion. Lloyd-Luke,London.

Brzezinski, A., Sadovsky, E., and Shafrir, E. (1961). Amer. 7. Obstet.Gynec., 82, 800.

Dancis, J., Lind, J., Oratz, M., Smolens, J., and Vara, P. (1961). Ibid.,82, 167.

Davies, J. (1960). Survey of Research in Gestation and the Developmen-tal Sciences. Williams and Wilkins, Baltimore.

Derrington, M. M., and Soothill, J. F. (1961). 7. Obstet. Gynaec. Brit.Cwlth, 68, 755.

Fairweather, D. V. I., Murray, S., Parkin, D., and Walker, W. (1963).Lancet, 2, 1190.and Walker, W. (1964). 7. Obstet. Gynaec. Brit. Cwlth, 71, 48.

Gairdner, D., Lawrie, N. R., and Hutcheon, M. (1950). Lancet, 2, 541.Goodlin, R. C. (1962). West. 7. Surg., 70, 181.Halvorsen, S., and Finne, P. H. (1963). Brit. med. 7., 1, 1132.Hibbard, B. M. (1963). Lancet, 2, 642.

Hyslop, R. S., and Whiley, W. (1960). Med. 7. Aust., 1, 622.Lawler, S. D., and Shatwell, H. S. (1962). Vox Sang., 7, 488.Liley, A. W. (1961). Amer. 7. Obstet. Gynec., 82, 1359.

(1963). Ibid., 86, 485.McKay, D. G., Richardson, M. V., and Hartig, A. T. (1958). Ibid., 75,

699.Mackay, E. V. (1961). Aust. N.Z. 7. Obstet. Gynaec., 1, 78.Mayer, M., Gueritat, P., Ducas, P., and Lewi, S. (1961). Presse med.,

69, 2493.Mollison, P. L. (1956). Blood Transfusion in Clinical Medicine, p. 478.

Blackwell, Oxford.Oram, W. J. (1960). N.Z. med. 7., 59, 255.Papadopoulos, N. M., Hess, W. C., O'Doherty, D., and McLane, J. E.

(1959). Clin. Chem., 5, 569.Parkin, D., Murray, S., and Walker, W. (1961). International Investiga-

tion of Rh Antibody Titrations: report to participants.Pickles, M. M. (1949). Haemolytic Disease of the Newborn. Blackwell,

Oxford.Stempfel, R., and Zetterstrdm, R. (1955). Pediatrics, 16, 184.Viergiver, E., Stroup, P. E., Sheff, M. F., and Westphal, M. (1962).

Obstet. and Gynec., 19, 664.Villee, C. A. (1957). Gestation. Transactions of 4th Conference Josiah

Macy Jr. Foundation, New York.Walker, A. H. C. (1957). Brit. med. 7., 2, 376.

and Jennison, R. F. (1962). Ibid., 2, 1152.Walker, W. (1960). Proc. 7th Congr. Europ. Soc. Haemat. London,

1959, Part II, p. 1186.and Bailey, B. M. (1956). 7. clin. Path., 9, 52.and Murray, S. (1956). Brit. med. 7., 1, 187.

and Russell, J. K. (1957). Lancet, 1, 348.White, D., Haidar, G. A., and Reinhold, J. G. (1958). Clin. Chem., 4,

211.Wild, A. E. (1961). Clin. Sci., 21, 221.Witebsky, E. (1948). Blood, Special Issue No. 2, p. 66.

and Mohn, J. F. (1945). 7. exp. Med., 82, 143.Zipursky, A., Pollock, J., Chown, B., and Israels, L. G. (1963). Lancet,

2, 493.

Examination of Amniotic Fluid in Rhesus Isoimmunization

JOHN G. ROBERTSON,* M.D., F.R.C.S.ED., M.R.C.O.G,

Brit. med. J., 1964, 2, 147-151

It is of some importance that the obstetrician who is supervisingthe pregnancy of a patient with rhesus isoimmunization shouldbe able to determine the severity of haemolytic disease affectingthe infant in utero. At the present time the induction of labourbefore the expected date of delivery is the only measure whichcan be routinely offered to such a patient in an attempt toprevent the death of the foetus.

Walker and Neligan (1955) considered that 98% of infantsborn alive with haemolytic disease should survive if the treat-ment they received was adequate, and in 1958 Walker statedthat with adequate care the neonatal mortality should be aslow as 6 per 1,000. Unfortunately, a large number of stillbirthsoccur each year from haemolytic disease-500 in England andWales (Lancet, 1958)-and in their series Walker and Murray(1956) reported a stillbirth rate of 10-15%. Tragically, 7 to80/ of pregnancies in which antibodies had been detected forthe first time ended in stillbirth, and it is primarily to preventthese stillbirths that prediction of the severity of the conditionis required.The methods of prediction available include the absolute

level of the antibody titre in the maternal blood and thechanges in this level, the previous history of the patient withregard to haemolytic disease in her children, the blood groupand rhesus genotype of the patient's husband, and radiologicalexamination. As none of these methods was entirely satis-factory, Bevis (1956) investigated the amount of " bloodpigment " present in the liquor amnii, obtaining this liquor byinserting a needle through the anterior abdominal wall (Bevis,1952). The liquor was then centrifuged and examined in a

spectrophotometer, the optical density being plotted againstthe wavelength at intervals between 360 and 700 m~i. Thepresence of a " bulge " from the expected " straight line " curvebetween 400 and 500 m~t indicated whether or not the foetuswas affected by haemolytic disease.

Walker (1957) used this test on 74 patients, and his predictionwas correct in 91 % of cases provided the test was carried outat 34 weeks and no distinction was made between " unaffected "and " mildly affected " infants. The actual severity of thecondition was not predicted, nor was the management stated.Liley (1960) described the results obtained in 200 tests andreported the onset of premature labour in seven patients, andalso two foetal deaths as a result of uterine infection. Cary(1960), McBride (1961), and Robertson (1961) were allimpressed by the ease of " amniocentesis," and considered theprocedure to be useful in patients with rhesus isoimmunization.Macbeth and Robertson (1961) found that the height of theoptical density indicated the severity of the haemolytic diseasein the foetus and indicated the optimum time for the induc-tion of labour. Liley (1961) also was able to predict theseverity.Mackay (1961) performed one test on each of 233 patients

and examined the liquor for " bilirubin staining." This deter-mined the time of induction of labour. Walker and Jennison(1962) reported a further series of 156 patients with a 91%correct prediction rate. They did not find that chemicalestimation of the bilirubin content of -the liquor was satisfactory.During the five years 1958-62 425 patients with rhesus

isoimmunization attended the Simpson Memorial MaternityPavilion of the Royal Infirmary of Edinburgh. Amniotic fluidwas obtained on at least one occasion from 252 of these patients,and the results obtained are reported.

* Senior Registrar, Simpson Memorial Maternity Pavilion, Royal Infirm-ary, Edinburgh; Lecturer, Department of Obstetrics and Gynae-cology, University of Edinburgh.

ID

on 18 April 2020 by guest. P

rotected by copyright.http://w

ww

.bmj.com

/B

r Med J: first published as 10.1136/bm

j.2.5402.147 on 18 July 1964. Dow

nloaded from

Treatment and Testing of Liquor Amnii

Liquor amnii was obtained between 33 and 35 weeks' gesta-tion, whenever possible. A lumbar-puncture needle was insertedinto the uterus transabdominally, as described by Bevis (1952).Sedation was not found to be necessary, but the procedure was

explained to the patient and 5 ml. of 10/ Duncaine solutionwas injected into the skin and subcutaneous tissues at theproposed site of puncture. After insertion of the needle andthe withdrawal of 20 ml. of fluid the patient was allowed home.The liquor was centrifuged slowly at first and then more

rapidly for 20 minutes, and then tested in a Unicam SP500 spectrophotometer. Readings of optical density were takenat 10-mne intervals between 360 and 600 m.t and then every25 m~t to 700 m~t, and the results plotted on semilogarithmicgraph paper.

As " false-positive " results were sometimes obtained (a bulgewhen the infant was unaffected) an additional test has beencarried out on many occasions since 1960.

Complications of Paracentesis Uteri

Among the complications reported after paracentesis uteriare uterine infection, the onset of premature labour, penetrationor separation of the placenta, and foetal death. In the presentseries there was no evidence of uterine infection and in no

case did labour begin spontaneously within two weeks of theparacentesis, provided this had been performed before 35 weeks.On a number of occasions the placenta was penetrated, causingblood-staining of the liquor, but in all cases the foetal heartvas heard after the procedure and no foetal deaths could beattributed to it. (Since these results were collected one foetushas died after paracentesis, blood being obtained during theprocedure.)

Diazo Test

Reagents.-(1) Diazo solution: (a) sulphanilic acid, 1 g./1,000 ml. ; hydrochloric acid, 15 ml./1,000 ml., water; (b)sodium nitrite, 0.5 g./100 ml. The solutions are freshly madeeach week, a solution for testing being prepared daily. This ismade by adding 0.3 ml. of solution b to 10 ml. of solution a.

(2) methyl alcohol.Method.-After the liquor has been centrifuged the following

solutions are added to a test-tube marked " total ": 2 ml. liquoramnii, 1 ml. diazo solution, and 3 ml. methyl alcohol. A secondtube is marked " direct " and contains: 2 ml. liquor amnii,1 ml. diazo solution, and 3 ml. hydrochloric acid. The solu-tions are allowed to stand for 10 minutes after preparation.Five solutions are tested with a water blank: (I) Untreatedliquor amnii; (II) Liquor with diazo solution and hydrochloricacid-direct bilirubin; (III) liquor with 4 ml. hydrochloricacid-control for No. II; (IV) liquor with diazo solution andmethyl alcohol-total bilirubin ; and (V) liquor with 1 ml.hydrochloric acid and methyl alcohol-control for No. IV.

Assessment of Severity

There seems to be no entirely satisfactory means of assessingthe severity of haemolytic disease, but the following classifica-tion, while not ideal, provides a reasonably practical basis.

Unaffected.-The rhesus group of the infant is the same as thatof the mother in all respects, or, if different, the direct Coombs testperformed on the infant's blood is negative and the infant isclinically unaffected.

Mildly affected.-The rhesus group of the infant differs fromthat of the mother and the direct Coombs test is positive, or there isclinical evidence of haemolytic disease, but no treatment-exchangeor simple transfusion-is required.

BRITISHMEDICAL JOURNAL

Moderately affected.-The infant is affected by haemolytic diseaseand requires treatment-that is, exchange transfusion-but is not asseverely affected as in the " severe " group.

Severely affected.-The infant is serologically affected with avery low cord haemoglobin (under 8.7 g./100 ml.)-anaemia group-or the indirect serum bilirubin rises rapidly, usually more thanonce, to 20 mg./100 ml. or more, despite exchange transfusions-hyperbilirubinaemia group. The infant survives.

Rhesus deaths.-The foetus is stillborn and haemolytic diseaseis confirmed by post-mortem examination as being the cause of deathor a very important factor ; or the child dies in the neonatal periodfrom haemolytic disease or from a cause directly associated withexchange transfusion.

It is difficult to devise a scheme for predicting the severitywhich takes into account the patient's maturity at delivery, thefoetal birth weight, the foetal haemoglobin at birth and sub-sequently, and changes in the level of serum bilirubin. Thescheme described is based on the treatment required by theinfant, and although the criteria for exchange transfusion vary

from hospital to hospital they are fairly constant in the same

hospital.

Determination of Severity from Optical-density Graphs

The severity of haemolytic disease is predicted by examiningthe optical density between 400 and 500 myt, particularly at450 m~l. Typical graphs are shown in Fig. 1.

(a) UNAFFECTED

0-4

(b) MODERATELY AFFECTED -0-5

700 600 500 400 700 3600 500 400WAVELENGTH IN my.

d(c SEVERELY AFFECTED

I ~~ ~~~~700 600 S00 400 700 600

WAVELENGTH IN mp.

-02

0-1-008-007-006-005-0-04 0

0-05 1-

0-

- 0-4 -'c

500 400

FIG. 1.-Spectrophotometric curves of untreated liquor specimens.

Optical-density Reading at 450 mbu.-This is not an accuratemeasure of the degree of involvement of the foetus, as thereis a very wide range for all degrees of severity as previouslydefined. In the present series the optical-density range for the" unaffected and mildly affected " group (considered together,as no special treatment is required and it is difficult to differen-tiate between the two groups) is from 0.075 to 0.32, with a meanof 0.178. The range for moderately affected foetuses is 0.066to 0.37 with a mean of 0.186. The difference between thesemeans is not statistically significant. The range for severelyaffected foetuses is from 0.094 to 0.78, with a mean of 0.292.The difference between the means of the moderate and severe

groups is highly significant (P<0.001). As the range is very

wide, and as there is considerable overlap among the groups,the measurement is not of prognostic value.

148 18 July 1964 Rhesus Isoimmunization-Robertson

I

on 18 April 2020 by guest. P

rotected by copyright.http://w

ww

.bmj.com

/B

r Med J: first published as 10.1136/bm

j.2.5402.147 on 18 July 1964. Dow

nloaded from

Height of Optical Density at 450 my% above the Baseline.-The baseline (the broken lines in Fig. 1) assumes that anunaffected curve is present and that the graph shows a" straight-line " curve. It is obtained by joining the points onthe graph obtained for the optical density at 400 mre and 500myL. Sometimes a more natural line is obtained by taking thebaseline from points before 400 mnu and after 500 myt. Inunaffected or mildly affected foetuses the difference betweenthe optical density on the graph and on the baseline at 450 m,>ranges between 0 and 0.02, with a mean of 0.016. Formoderately affected foetuses the range is 0.02 to 0.07, with a

mean of 0.047, and the difference between these means is highlysignificant (P<0.001). In severely affected foetuses the rangeof optical-density height above the baseline is considerable-between 0.07 and 0.7, with a mean of 0.154-and the differencebetween the means of the moderately affected and severelyaffected groups is highly significant (P<0.001). The greaterthe height of optical density above the baseline the more severe

the haemolytic disease as a general rule.Height of Optical Density above the Baseline in cm. at 450

mju.-The optical density is plotted on semilogarithmic graphpaper and the actual reading of optical density may be high or

low on the scale. For this reason the height recorded abovethe baseline may be proportionately greater or less than theactual measured optical-density difference. This applies parti-cularly below 0.1, where a height in centimetres is proportion-ately greater than the difference in optical density. The rangefor unaffected and mildly affected foetuses is between 0 and 1cm., with a mean of 0.6 cm.; for the moderately affected foetusesthe range is 1-2 cm., with a mean of 1.62 cm.; and for theseverely affected foetuses the height is greater than 2 cm., themean being 3 cm. The difference between the meansof the unaffected and mildly affected and the moderatelyaffected groups is highly significant (P<0.001) and similarly,the difference between the means of the moderately and severelyaffected groups is highly significant (P<0.001).

False-positive and False-negative Predictions

An infant was thought to be affected, but.was unaffected, on42 occasions (false-positives). On eight of these the infant was

jaundiced in the neonatal period with high levels of serumbilirubin but no evidence of haemolytic disease due to rhesusincompatibility. In four patients the period of gestation at thetime of the test was greater than 35 weeks, and in one the dategiven was probably incorrect. On three occasions, early in theseries, the graphs were difficult to interpret and an incorrectprediction was made.

Since 1960 the diazo test has been used in an attempt todetect the presence of appreciable amounts of indirect bilirubin,

700 600 500 460WAVELENGTH IN mp.

FIG. 2.-Liquor curves after treatmentwith diazo solutions, and controls. Lessaffected than suggested by untreatedliquor. I= Untreated liquor. II=Liquor+ diazo. III = Liquor+ hydro-chloric acid. IV= Liquor + diazo+methyl alcohol. V= Liquor + hydro-

chloric acid + methyl alcohoL

SEVERELY AFFECTED ,

700 600 500 4600WAVELENGTH IN

BRMSTOMEDICAL JOURNAL 149

as it was thought that this might eliminate false-positive results.It was sometimes found that the graph of the untreated liquorshowed a definite bulge between 400 and 500 m~u, but thatthis was associated with direct bilirubin only, as is illustratedin Fig. 2. The bulge between 500 and 600 mix, with a peakat 550 mpt in curve II, indicated direct bilirubin and, as thisis minimal in curve IV, it suggests that very little indirectbilirubin is present and that the infant is unaffected. Thiswas found on 21 occasions in 1961 and 1962.On a few occasions the bulge in curve IV has been much

greater than expected and has suggested that the infant wasmore severely affected than calculated from the untreated liquor-curve I. This is illustrated in Fig. 3.

Eleven patients were thought to have unaffected infants,although this was subsequently proved wrong. In 10 of thesethe test was performed after 36 weeks owing to errors in calcu-lation on the part of the patients, while in one no explanationwas found.

Results

The results obtained are recorded in Table I. Because ofthe difficulty in interpreting the curve after 35 weeks' gestation,these " late tests " are considered separately. The number of" incorrect" predictions corrected by the diazo test are alsoshown. (The test has been used whenever possible since 1960.)

TABLE I.-Results of Examination of Liquor Amnii with Severity ofHaemolytic Disease of Infant

Tested before Tested after Cor-35 Weeks 35 Weeks rected

Foetal Results Total byCorrect In- Correct In- Diazo

correct correct Test*Unaffected or mildly

affected .. 37 38 12 4 91 21Moderately affected .. 73 7 12 3 95 1Severely .... 44 1 4 0 49 0Deaths from haemolytic

disease .. .. 9 5 2 1 17 1

Total .. 163 51 30 8 252 23

* Correction by diazo ignores the direct bilirubin in the liquor.

In Table II the " incorrect " predictions are studied in moredetail. In Table III the correct prediction rate in the seriesis shown to be 86%, while in 1961-62, when the diazo test wasused, the predictions were correct in 91 % of cases. The 69%correct prediction rate in the second time-period, without thediazo test, is probably the result of more exact interpretationof the graphs during this time, with a more definite standardi-zation of the indications for treatment of the infant (largelyaffecting the mildly affected and moderately affected groups).Also, the knowledge that the diazo test had been performed

may have influenced the prediction based on theuntreated liquor amnii.

05-

-04

-05

-0-10-008.- 007006

,- . *

-0050-04

-I

C,

z

In

0-03

FIG. 3.-Liquor curves after treatmentwith diazo solutions, and controls.More severely affected than suggested byuntreated liquor. Explanation of curves

as in Fig. 2.

Comparison of Antibody Titres and Liquor Analysisin Predicting Severity.-It is difficult to assess, in retro-spect, the prediction made by studying the antibodytitres, but it seems that although the liquor predictionhas been correct on 60 occasions when the titres were

wrong, the titres were correct 29 times when the liquorwas incorrect (the diazo test not being used).

Observations on Liquor Curves of Severely Affected.-No correlation between the optical density and eitherthe haemoglobin or serum bilirubin of the infant atbirth could be found, possibly because of the timeelapsing between the test and the delivery of the infant.In the severely affected group the curves were foundto have certain characteristic features. In the hyper-bilirubinaemia group a plateau was found between 410and 460 mu in many cases, while in nine others a

marked hump-peak was found at 410 mMu, in additionto the plateau (Fig. 4). In the anaemia group there

18 July 1964 Rhesus Isoimmunization-Robertson

on 18 April 2020 by guest. P

rotected by copyright.http://w

ww

.bmj.com

/B

r Med J: first published as 10.1136/bm

j.2.5402.147 on 18 July 1964. Dow

nloaded from

150 18 July 1964 Rhesus Isoimmunization-Robertson

Incorrect Predictions

No. Prediction Made

42 38 moderate. 4 severe10 10 unaffected1 16 6 moderate

59

TABLE III.-Results of Predictions for Two Time-periods, the DiazoTest Being Used in the Second Period

Foetal Results

Unaffected andmildly affected

Moderately affectedSeverely affected

incl. rhesus deaths

Total . .

Prediction Correct

1958-60 1961-2

No. % No.

33 66 16 (37) 39 (90)56 90 29 (30) 88 (91)

33 89 26 (27) 90 (93)122_8271(94) 69 (91122 82 71 (94) 69 (91)

I_

Total

No. °o

(70) (77)

85 (86) 90 (91)

59 (60) 89 (91)

193 (216) 77 (86)

Figures in parentheses represent results after diazo correction.

were no hump-peaks, but two-thirds of the patients hadplateaux (Fig. 5). In nine out of 13 infants dying from haemo-lytic disease humps and plateaux were found.

0 4

s0

0 004l , , 0-2 . , 0-

0-08~~r-

~~~~0-07 -.

600 500 4000 700 600 500 0 0080*05 0-07

0-04 -0-06

0.03 0.04

0-02 0-03

1700 60of0 500 400 700 600 500 400

WAVELENGTH IN my. WAVELENGTH IN my.FIG. 4 FIG. 5

FIG. 4.-Severely affected-" hyperbilirubinaemia," with plateau and

hump-peak. FIG. 5.-Severely affected-" anaemia," with plateau.

Management of Patients

After examination of the liquor all the information relevant

to the patient's case was studied and a decision regarding the

time of induction and the method of delivery made. Unaffected

and mildly affected infants were delivered by the expected date

of delivery. If a moderately affected. infant was predictedlabour was induced between 37 and 39 weeks, while if the infant

was considered to be severely affected it was delivered between

35 and 37 weeks. Occasionally " heroic" measures were called

for before 35 weeks' gestation. If induction was decided upon

it was necessary to set a time limit for delivery, and if the induc-

tion failed caesarean section was required. Labour was usually

induced by artificial rupture of the membranes and oxytocin

drip when necessary.Of the 252 infants delivered after examination of the liquor,

17 died as a result of haemolytic disease. These included two

neonatal deaths. Four other infants were affected by haemoly-tic disease of the newborn but died of other conditions, while

two of the infants lost were unaffected. The total foetal

mortality in the series was 9%, and the mortality rate from

haemolytic disease was 700.

Discussion

The most logical method of obtaining information about a

foetus in utero during pregnancy is to examine the foetal tissues

themselves or some fluid in close proximity to the foetus. The

liquor amnii is such a fluid, and any "excretion products"from the foetus are likely to be present in it. If there is

haemolysis of the foetal blood by the antibodies, breakdown

products will be directly related to the amount of haemolysis,and so it is possible to predict whether the infant is affected or

not by haemolytic disease and to estimate the severity of this

disease.

The tests are best carried out at 34 weeks, and if blood is

routinely tested at 32 weeks from patients who may developantibodies-those who are rhesus-negative or rhesus-positive

having had blood-transfusions, or whose previous infants have

had unexplained jaundice-the presence of immunization will

be detected in time to arrange for a liquor test. This should

be carried out in all patients found to have antibodies. Althoughthe procedure is not difficult, it is made easier by frequent

performance. The haemolytic process may become more

marked as term approaches, and the degree of severity may be

greater than anticipated, but this can usually be allowed for in

making the prediction. The prediction of severity is not

quantitative but merely gives some indication of the best time to

induce labour in order to avoid an intrauterine death or a very

severely affected infant. It is essential that efforts are made to

reduce the mortality from haemolytic disease, and as this is

primarily due to stillbirths, early induction is the only practical

therapeutic method available at the present time. Neonatal

deaths are rare and with adequate paediatric supervision should

continue to be so.

There are other causes of haemolysis in the foetus in utero

or after delivery, including drugs taken in the antenatal period.These may give rise to false-positive predictions, as may the

presence of direct bilirubin in the liquor. For this reason a

diazo test was devised for estimating the amount of indirect

bilirubin. A number of incorrect predictions during the last

three years were made as a result of testing untreated liquor,but these were corrected by the diazo test.

The difference between the optical density recorded at 450

mi' and the baseline recording at the same wavelength gave a

range for the different degrees of severity, especially when this

was supplemented by the difference measured in centimetres

between the two readings. Although there was some overlap,the results were accurate enough to indicate the optimum time

for induction to the nearest two or three days. The readingsobtained could not be correlated with the birth haemoglobin or

serum-bilirubin concentration, probably because of the variable

time intervals between the test and delivery. If intrauterine

transfusion for the prevention of foetal death from hydropsfoetalis becomes a practical procedure (Liley, 1963), liquorexamination will be of value in determining those infants for

whom transfusion might be life-saving. If a bulge suggestiveof a severely anaemic foetus is obtained transfusion might helpto keep it alive until premature delivery can be effected.

Greater accuracy in deciding suitable conditions for transfusion,

however, would be obtained if foetal blood was aspirated and

the haemoglobin level determined.

Routine induction of labour at gestations between 35 and

40 weeks has been advocated by various writers (Aaro, 1959;

Jacobs, 1959 ; Tovey and Valaes, 1959 ; Wiener, 1959 ; Browne,

1960; Goplerud, 1961 ; Townsend et al., 1961), and it is

generally agreed that immunized patients should be delivered

before term. Any procedure which allows a rational timing of

induction is of value, and liquor examination would seem to be

such a procedure.

Summary

The method of obtaining and testing the amniotic fluid in

patients with rhesus isoimmunization is described, with

particular reference to a diazo test which is used to determine

the presence of indirect bilirubin.

TABLE II.-Predictions and Foetal Results in " Incorrect Group "

Foetal Results

Unaffected or mildly affected ..Moderately affectedSeverely ,.Deaths from haemolytic disease ..

Total . . .

BRTSHMEDICAL JOURNAL-

on 18 April 2020 by guest. P

rotected by copyright.http://w

ww

.bmj.com

/B

r Med J: first published as 10.1136/bm

j.2.5402.147 on 18 July 1964. Dow

nloaded from

18 July 1964 Rhesus Isoimmunization-Robertson MEDICBALJOURNAL 151

The optical density at 450 mju above the baseline is used topredict the severity of haemolytic disease in the foetus, and 193out of 252 predictions were correct. The diazo test gave thecorrect result in a further 23 patients. Seventeen infants werelost as a result of haemolytic disease (7%).

It is concluded that liquor examination is of definite valuein the prediction of severity of haemolytic disease, and indetermining the subsequent management of the patients.

I am grateful to Professor R. J. Kellar and Dr. G. D. Matthewfor their help and encouragement during this investigation. I amalso grateful to the consultant obstetricians in the South-EasternRegion of Scotland for permitting me to obtain liquor from theirpatients. I also wish to express my thanks to Miss J. Cameron,A.M.I.L.T., for her assistance in devising the diazo liquor test; tothe technicians in the clinical laboratory of the Simpson MaternityHospital for testing the liquor; and to Sister M. Isdale, of theSimpson Maternity Hospital, for her help in carrying out the testsand in the investigation of the patients. The co-operation of Dr.R. A. Cumming and the staff of the Regional Blood TransfusionCentre and of the paediatricians attached to the Simpson Maternity

Hospital has been invaluable. This material forms part of a thesisfor the degree of Doctor of Medicine accepted by the University ofEdinburgh.

REFERENCES

Aaro, L. A. (1959). Surg. Clin. N. Amer., 39, 1125.Bevis, D. C. A. (1952). Lancet, 1, 395.

(1956). 7. Obstet. Gynaec. Brit. Emp., 63, 68.Browne, J. C. McC. (1960). Ibid., 67, 746.Cary, W. (1960). Med. 7. Aust., 2, 778.Goplerud, C. P. (1961). Obstet. and Gynec., 17, 355.Jacobs, W. H. (1959). Ibid., 13, 314.Lancet, 1958, 2, 303.Liley, A. W. (1960). N.Z. med. 7., 59, 581.

(1961). Amer. 7. Obstet. Gynec., 82, 1359.(1963). Brit. med. 7., 2, 1107.

Macbeth, R. D., and Robertson, S. B. (1961). Med. 7. Aust., 2, 573.McBride, W. G. (1961). Ibid., 1, 403.Mackay, E. V. (1961). Aust. N.Z. 7. Obstet. Gynaec., 1, 78.Robertson, S. E. J. (1961). Med. 7. Aust., 1, 398.Tovey, G. H., and Valaes, T. (1959). Lancet, 2, 521.Townsend, S. L., Mackay, E. V. Shelton, J. G., Krieger, V. I., and

Campbell, K. I. (1961). 7. Obstet. Gynaec. Brit. Cwlth, 68, 382.Walker, A. H. C. (1957). Brit. med. 7., 2, 376.- and Jennison, R. F. (1962). Ibid., 2, 1152.Walker, W. (1958). Vox Sang (Basel), 3, 225.- and Murray, S. (1956). Brnt. med. 7., 1, 187.

and Neligan, G. A. (1955). Ibid., 1, 681.Wiener, A. S. (1959). Exp. Med. Sung., 17, 15.

Hand Eczema

F. RAY BETTLEY,* M.D., F.R.C.P.

Brit. med. 7., 1964, 2, 151-155

When eczema affects the hands it tends to be disabling as wellas uncomfortable; to the patient it is ever before him, inescap-able; it embarrasses him because it is so often easily visible toothers; and it is common.The 106 cases of hand eczema which form the basis of this

report were in patients seen by me for the first time over aperiod of 12 months and represent, therefore, about 5% ofall patients attending the dermatological out-patient clinic.Even so, they are selected, since they are only those cases wherethe hands were originally involved and in which the handlesion was the primary complaint. Other eczemas, where thehands were later involved as a part of a more widespreaderuption, have not been included and would, no doubt, haveprovided a further substantial number. All the patients inthe present series were seen, investigated, and treated by mepersonally with the main object of forming a consistent assess-ment of causative factors and of the value of treatment. Finalreview was carried out 18 to 24 months after first attendance;seven failed to report for follow-up.The group contained 56 men and 50 women, a slight

difference of no likely significance. The age at onset (Table I)TABLE I.-Age at Onset of Hand Eczema

Age (yrs) -5 -10 -15 -20 -25 -30 -35 -40 -45 -50 -55 -60 -65 -70 TotalMen .1 1 - 3 6 3 4 7 4 11 7 4 2 3 56Women .. 2 3 5 7 4 4 5 4 7 5 2 2 - 50

shows a difference in the sexes. In men a peak incidence isfound at 46-50 years of age, while in women the age distribu-tion is more level from 16 years onwards; this might beattributed to the effect of housework with its cleansers andprimary irritants, but for reasons given below this suggestioncannot be substantiated.* The Middlesex Hospital and St. John's Hospital for Diseases of the

Skin, London.

Clinical Groups

Among the whole series it is possible to define certain clinicalgroups.

Nummular Eczema

The term " nummular " is applied to a type of discoideczema, usually affecting the backs of the hands and fingers,sometimes the forearms, in which the eczematous areas arewell outlined with normal skin between. There were 12 menand four women showing this pattern, one case being possiblyatopic. None gave positive reactions to routine patch tests.

This group was distinguished by its relatively short duration,often clearing up in the course of some months.

Nickel Allergy

Calnan and Wells (1956) drew attention to the fact that nickelis a very common cause of epidermal allergy in women in theUnited Kingdom. They pointed out, too, that these patientsare prone to hand eczema even after nickel dermatitis has healedand without identifiable continuing contact with nickel. Sixwomen in the present series were of this kind. All of themknew of their proneness to metal rashes and avoided contactwith metals. The duration of their hand eczema was fromone month to three years, but nickel sensitivity was of manyyears' duration.

Nickel sensitivity in men is more often occupational,occurring, for example, in electroplaters. Two men in thepresent group were found on patch-testing to be nickel-sensitive; one of these gave a history of a rash beneath a wrist-

E

on 18 April 2020 by guest. P

rotected by copyright.http://w

ww

.bmj.com

/B

r Med J: first published as 10.1136/bm

j.2.5402.147 on 18 July 1964. Dow

nloaded from